opencv/3rdparty/libtiff/tif_lzma.c
2012-10-17 15:32:23 +04:00

496 lines
15 KiB
C

/* $Id: tif_lzma.c,v 1.4 2011-12-22 00:29:29 bfriesen Exp $ */
/*
* Copyright (c) 2010, Andrey Kiselev <dron@ak4719.spb.edu>
*
* Permission to use, copy, modify, distribute, and sell this software and
* its documentation for any purpose is hereby granted without fee, provided
* that (i) the above copyright notices and this permission notice appear in
* all copies of the software and related documentation, and (ii) the names of
* Sam Leffler and Silicon Graphics may not be used in any advertising or
* publicity relating to the software without the specific, prior written
* permission of Sam Leffler and Silicon Graphics.
*
* THE SOFTWARE IS PROVIDED "AS-IS" AND WITHOUT WARRANTY OF ANY KIND,
* EXPRESS, IMPLIED OR OTHERWISE, INCLUDING WITHOUT LIMITATION, ANY
* WARRANTY OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
*
* IN NO EVENT SHALL SAM LEFFLER OR SILICON GRAPHICS BE LIABLE FOR
* ANY SPECIAL, INCIDENTAL, INDIRECT OR CONSEQUENTIAL DAMAGES OF ANY KIND,
* OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS,
* WHETHER OR NOT ADVISED OF THE POSSIBILITY OF DAMAGE, AND ON ANY THEORY OF
* LIABILITY, ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE
* OF THIS SOFTWARE.
*/
#include "tiffiop.h"
#ifdef LZMA_SUPPORT
/*
* TIFF Library.
*
* LZMA2 Compression Support
*
* You need an LZMA2 SDK to link with. See http://tukaani.org/xz/ for details.
*
* The codec is derived from ZLIB codec (tif_zip.c).
*/
#include "tif_predict.h"
#include "lzma.h"
#include <stdio.h>
/*
* State block for each open TIFF file using LZMA2 compression/decompression.
*/
typedef struct {
TIFFPredictorState predict;
lzma_stream stream;
lzma_filter filters[LZMA_FILTERS_MAX + 1];
lzma_options_delta opt_delta; /* delta filter options */
lzma_options_lzma opt_lzma; /* LZMA2 filter options */
int preset; /* compression level */
lzma_check check; /* type of the integrity check */
int state; /* state flags */
#define LSTATE_INIT_DECODE 0x01
#define LSTATE_INIT_ENCODE 0x02
TIFFVGetMethod vgetparent; /* super-class method */
TIFFVSetMethod vsetparent; /* super-class method */
} LZMAState;
#define LState(tif) ((LZMAState*) (tif)->tif_data)
#define DecoderState(tif) LState(tif)
#define EncoderState(tif) LState(tif)
static int LZMAEncode(TIFF* tif, uint8* bp, tmsize_t cc, uint16 s);
static int LZMADecode(TIFF* tif, uint8* op, tmsize_t occ, uint16 s);
static const char *
LZMAStrerror(lzma_ret ret)
{
switch (ret) {
case LZMA_OK:
return "operation completed successfully";
case LZMA_STREAM_END:
return "end of stream was reached";
case LZMA_NO_CHECK:
return "input stream has no integrity check";
case LZMA_UNSUPPORTED_CHECK:
return "cannot calculate the integrity check";
case LZMA_GET_CHECK:
return "integrity check type is now available";
case LZMA_MEM_ERROR:
return "cannot allocate memory";
case LZMA_MEMLIMIT_ERROR:
return "memory usage limit was reached";
case LZMA_FORMAT_ERROR:
return "file format not recognized";
case LZMA_OPTIONS_ERROR:
return "invalid or unsupported options";
case LZMA_DATA_ERROR:
return "data is corrupt";
case LZMA_BUF_ERROR:
return "no progress is possible (stream is truncated or corrupt)";
case LZMA_PROG_ERROR:
return "programming error";
default:
return "unindentified liblzma error";
}
}
static int
LZMAFixupTags(TIFF* tif)
{
(void) tif;
return 1;
}
static int
LZMASetupDecode(TIFF* tif)
{
LZMAState* sp = DecoderState(tif);
assert(sp != NULL);
/* if we were last encoding, terminate this mode */
if (sp->state & LSTATE_INIT_ENCODE) {
lzma_end(&sp->stream);
sp->state = 0;
}
sp->state |= LSTATE_INIT_DECODE;
return 1;
}
/*
* Setup state for decoding a strip.
*/
static int
LZMAPreDecode(TIFF* tif, uint16 s)
{
static const char module[] = "LZMAPreDecode";
LZMAState* sp = DecoderState(tif);
lzma_ret ret;
(void) s;
assert(sp != NULL);
if( (sp->state & LSTATE_INIT_DECODE) == 0 )
tif->tif_setupdecode(tif);
sp->stream.next_in = tif->tif_rawdata;
sp->stream.avail_in = (size_t) tif->tif_rawcc;
if ((tmsize_t)sp->stream.avail_in != tif->tif_rawcc) {
TIFFErrorExt(tif->tif_clientdata, module,
"Liblzma cannot deal with buffers this size");
return 0;
}
/*
* Disable memory limit when decoding. UINT64_MAX is a flag to disable
* the limit, we are passing (uint64_t)-1 which should be the same.
*/
ret = lzma_stream_decoder(&sp->stream, (uint64_t)-1, 0);
if (ret != LZMA_OK) {
TIFFErrorExt(tif->tif_clientdata, module,
"Error initializing the stream decoder, %s",
LZMAStrerror(ret));
return 0;
}
return 1;
}
static int
LZMADecode(TIFF* tif, uint8* op, tmsize_t occ, uint16 s)
{
static const char module[] = "LZMADecode";
LZMAState* sp = DecoderState(tif);
(void) s;
assert(sp != NULL);
assert(sp->state == LSTATE_INIT_DECODE);
sp->stream.next_in = tif->tif_rawcp;
sp->stream.avail_in = (size_t) tif->tif_rawcc;
sp->stream.next_out = op;
sp->stream.avail_out = (size_t) occ;
if ((tmsize_t)sp->stream.avail_out != occ) {
TIFFErrorExt(tif->tif_clientdata, module,
"Liblzma cannot deal with buffers this size");
return 0;
}
do {
/*
* Save the current stream state to properly recover from the
* decoding errors later.
*/
const uint8_t *next_in = sp->stream.next_in;
size_t avail_in = sp->stream.avail_in;
lzma_ret ret = lzma_code(&sp->stream, LZMA_RUN);
if (ret == LZMA_STREAM_END)
break;
if (ret == LZMA_MEMLIMIT_ERROR) {
lzma_ret r = lzma_stream_decoder(&sp->stream,
lzma_memusage(&sp->stream), 0);
if (r != LZMA_OK) {
TIFFErrorExt(tif->tif_clientdata, module,
"Error initializing the stream decoder, %s",
LZMAStrerror(r));
break;
}
sp->stream.next_in = next_in;
sp->stream.avail_in = avail_in;
continue;
}
if (ret != LZMA_OK) {
TIFFErrorExt(tif->tif_clientdata, module,
"Decoding error at scanline %lu, %s",
(unsigned long) tif->tif_row, LZMAStrerror(ret));
break;
}
} while (sp->stream.avail_out > 0);
if (sp->stream.avail_out != 0) {
TIFFErrorExt(tif->tif_clientdata, module,
"Not enough data at scanline %lu (short %lu bytes)",
(unsigned long) tif->tif_row, (unsigned long) sp->stream.avail_out);
return 0;
}
tif->tif_rawcp = (uint8 *)sp->stream.next_in; /* cast away const */
tif->tif_rawcc = sp->stream.avail_in;
return 1;
}
static int
LZMASetupEncode(TIFF* tif)
{
LZMAState* sp = EncoderState(tif);
assert(sp != NULL);
if (sp->state & LSTATE_INIT_DECODE) {
lzma_end(&sp->stream);
sp->state = 0;
}
sp->state |= LSTATE_INIT_ENCODE;
return 1;
}
/*
* Reset encoding state at the start of a strip.
*/
static int
LZMAPreEncode(TIFF* tif, uint16 s)
{
static const char module[] = "LZMAPreEncode";
LZMAState *sp = EncoderState(tif);
(void) s;
assert(sp != NULL);
if( sp->state != LSTATE_INIT_ENCODE )
tif->tif_setupencode(tif);
sp->stream.next_out = tif->tif_rawdata;
sp->stream.avail_out = (size_t)tif->tif_rawdatasize;
if ((tmsize_t)sp->stream.avail_out != tif->tif_rawdatasize) {
TIFFErrorExt(tif->tif_clientdata, module,
"Liblzma cannot deal with buffers this size");
return 0;
}
return (lzma_stream_encoder(&sp->stream, sp->filters, sp->check) == LZMA_OK);
}
/*
* Encode a chunk of pixels.
*/
static int
LZMAEncode(TIFF* tif, uint8* bp, tmsize_t cc, uint16 s)
{
static const char module[] = "LZMAEncode";
LZMAState *sp = EncoderState(tif);
assert(sp != NULL);
assert(sp->state == LSTATE_INIT_ENCODE);
(void) s;
sp->stream.next_in = bp;
sp->stream.avail_in = (size_t) cc;
if ((tmsize_t)sp->stream.avail_in != cc) {
TIFFErrorExt(tif->tif_clientdata, module,
"Liblzma cannot deal with buffers this size");
return 0;
}
do {
lzma_ret ret = lzma_code(&sp->stream, LZMA_RUN);
if (ret != LZMA_OK) {
TIFFErrorExt(tif->tif_clientdata, module,
"Encoding error at scanline %lu, %s",
(unsigned long) tif->tif_row, LZMAStrerror(ret));
return 0;
}
if (sp->stream.avail_out == 0) {
tif->tif_rawcc = tif->tif_rawdatasize;
TIFFFlushData1(tif);
sp->stream.next_out = tif->tif_rawdata;
sp->stream.avail_out = (size_t)tif->tif_rawdatasize; /* this is a safe typecast, as check is made already in LZMAPreEncode */
}
} while (sp->stream.avail_in > 0);
return 1;
}
/*
* Finish off an encoded strip by flushing the last
* string and tacking on an End Of Information code.
*/
static int
LZMAPostEncode(TIFF* tif)
{
static const char module[] = "LZMAPostEncode";
LZMAState *sp = EncoderState(tif);
lzma_ret ret;
sp->stream.avail_in = 0;
do {
ret = lzma_code(&sp->stream, LZMA_FINISH);
switch (ret) {
case LZMA_STREAM_END:
case LZMA_OK:
if ((tmsize_t)sp->stream.avail_out != tif->tif_rawdatasize) {
tif->tif_rawcc =
tif->tif_rawdatasize - sp->stream.avail_out;
TIFFFlushData1(tif);
sp->stream.next_out = tif->tif_rawdata;
sp->stream.avail_out = (size_t)tif->tif_rawdatasize; /* this is a safe typecast, as check is made already in ZIPPreEncode */
}
break;
default:
TIFFErrorExt(tif->tif_clientdata, module, "Liblzma error: %s",
LZMAStrerror(ret));
return 0;
}
} while (ret != LZMA_STREAM_END);
return 1;
}
static void
LZMACleanup(TIFF* tif)
{
LZMAState* sp = LState(tif);
assert(sp != 0);
(void)TIFFPredictorCleanup(tif);
tif->tif_tagmethods.vgetfield = sp->vgetparent;
tif->tif_tagmethods.vsetfield = sp->vsetparent;
if (sp->state) {
lzma_end(&sp->stream);
sp->state = 0;
}
_TIFFfree(sp);
tif->tif_data = NULL;
_TIFFSetDefaultCompressionState(tif);
}
static int
LZMAVSetField(TIFF* tif, uint32 tag, va_list ap)
{
static const char module[] = "LZMAVSetField";
LZMAState* sp = LState(tif);
switch (tag) {
case TIFFTAG_LZMAPRESET:
sp->preset = (int) va_arg(ap, int);
lzma_lzma_preset(&sp->opt_lzma, sp->preset);
if (sp->state & LSTATE_INIT_ENCODE) {
lzma_ret ret = lzma_stream_encoder(&sp->stream,
sp->filters,
sp->check);
if (ret != LZMA_OK) {
TIFFErrorExt(tif->tif_clientdata, module,
"Liblzma error: %s",
LZMAStrerror(ret));
}
}
return 1;
default:
return (*sp->vsetparent)(tif, tag, ap);
}
/*NOTREACHED*/
}
static int
LZMAVGetField(TIFF* tif, uint32 tag, va_list ap)
{
LZMAState* sp = LState(tif);
switch (tag) {
case TIFFTAG_LZMAPRESET:
*va_arg(ap, int*) = sp->preset;
break;
default:
return (*sp->vgetparent)(tif, tag, ap);
}
return 1;
}
static const TIFFField lzmaFields[] = {
{ TIFFTAG_LZMAPRESET, 0, 0, TIFF_ANY, 0, TIFF_SETGET_INT, TIFF_SETGET_UNDEFINED,
FIELD_PSEUDO, TRUE, FALSE, "LZMA2 Compression Preset", NULL },
};
int
TIFFInitLZMA(TIFF* tif, int scheme)
{
static const char module[] = "TIFFInitLZMA";
LZMAState* sp;
lzma_stream tmp_stream = LZMA_STREAM_INIT;
assert( scheme == COMPRESSION_LZMA );
/*
* Merge codec-specific tag information.
*/
if (!_TIFFMergeFields(tif, lzmaFields, TIFFArrayCount(lzmaFields))) {
TIFFErrorExt(tif->tif_clientdata, module,
"Merging LZMA2 codec-specific tags failed");
return 0;
}
/*
* Allocate state block so tag methods have storage to record values.
*/
tif->tif_data = (uint8*) _TIFFmalloc(sizeof(LZMAState));
if (tif->tif_data == NULL)
goto bad;
sp = LState(tif);
memcpy(&sp->stream, &tmp_stream, sizeof(lzma_stream));
/*
* Override parent get/set field methods.
*/
sp->vgetparent = tif->tif_tagmethods.vgetfield;
tif->tif_tagmethods.vgetfield = LZMAVGetField; /* hook for codec tags */
sp->vsetparent = tif->tif_tagmethods.vsetfield;
tif->tif_tagmethods.vsetfield = LZMAVSetField; /* hook for codec tags */
/* Default values for codec-specific fields */
sp->preset = LZMA_PRESET_DEFAULT; /* default comp. level */
sp->check = LZMA_CHECK_NONE;
sp->state = 0;
/* Data filters. So far we are using delta and LZMA2 filters only. */
sp->opt_delta.type = LZMA_DELTA_TYPE_BYTE;
/*
* The sample size in bytes seems to be reasonable distance for delta
* filter.
*/
sp->opt_delta.dist = (tif->tif_dir.td_bitspersample % 8) ?
1 : tif->tif_dir.td_bitspersample / 8;
sp->filters[0].id = LZMA_FILTER_DELTA;
sp->filters[0].options = &sp->opt_delta;
lzma_lzma_preset(&sp->opt_lzma, sp->preset);
sp->filters[1].id = LZMA_FILTER_LZMA2;
sp->filters[1].options = &sp->opt_lzma;
sp->filters[2].id = LZMA_VLI_UNKNOWN;
sp->filters[2].options = NULL;
/*
* Install codec methods.
*/
tif->tif_fixuptags = LZMAFixupTags;
tif->tif_setupdecode = LZMASetupDecode;
tif->tif_predecode = LZMAPreDecode;
tif->tif_decoderow = LZMADecode;
tif->tif_decodestrip = LZMADecode;
tif->tif_decodetile = LZMADecode;
tif->tif_setupencode = LZMASetupEncode;
tif->tif_preencode = LZMAPreEncode;
tif->tif_postencode = LZMAPostEncode;
tif->tif_encoderow = LZMAEncode;
tif->tif_encodestrip = LZMAEncode;
tif->tif_encodetile = LZMAEncode;
tif->tif_cleanup = LZMACleanup;
/*
* Setup predictor setup.
*/
(void) TIFFPredictorInit(tif);
return 1;
bad:
TIFFErrorExt(tif->tif_clientdata, module,
"No space for LZMA2 state block");
return 0;
}
#endif /* LZMA_SUPORT */
/* vim: set ts=8 sts=8 sw=8 noet: */